Japan Markets ViewIndustrial Production Index Forecasted with KDDI’s Human Flow Data (1)
Jul 14, 2022
Human flow data provided by KDDI, a major Japanese telecommunications company, could be used to calculate with higher accuracy than the forecasts published by the Ministry of Economy, Trade and Industry (METI) for some industries. QUICK, a financial information provider, revealed it with a multiple regression analysis. This is likely to be a new method of utilizing big data.
Big data analysis using alternative data has become popular in the financial and economic fields. It is widely used for stock prices, business conditions, and economic indicators. In this article, KDDI’s statistical data provision service on human flow, “KDDI Location Data” (hereafter referred to as “human flow data”) is used to confirm the correlation with the industrial production index by industry published monthly by METI.
The KDDI service can acquire data for areas where a company’s factories are located, with a minimum size of 10 meters squared mesh. The frequency of data updates is daily.
KDDI obtains GPS location information of mobile phone devices with the individual consent from customers and processes it into an anonymous format. In addition, the map locations are linked to the issue codes of listed companies, making it easy to use for analyzing financial and economic data.
Human Flow and Industrial Production Index Generally Correlated
This time, we targeted four specific industries: Transport Equipment, Production Machinery, Chemicals (excl. Medicine) and Fabricated Metals. Human flow data for factories of specific companies were compiled to produce daily human flow data for each industry sector. The graph below shows the monthly human flow data for each industry overlaid on the actual data for the industrial production index.
The results show that the months of increase and decrease coincided for all industries. Although there has been some impact from the spread of COVID-19 since the beginning of 2020, generally, the direction of the movement is consistent.
Transport Equipment
Production Machinery
Chemicals (excl. Medicine)
Fabricated Metals
METI does not disclose which companies are included in the industrial production index by industry. The following is a list of the components of each industry set by the analyst for this forecast.
Transport Equipment
DENSO (6902), Toyota Motor (7203), Subaru (7270), Shimano (7309), Suzuki Motor (7269), Toyota Industries (6201), Honda Motor (7267)
Production Machinery
Renesas Electronics (6723), Mitsubishi Heavy Industries (7011), DISCO (6146), Makita (6586), Kubota (6326), Daikin Industries (6367)
Chemicals (excl. Medicine)
Mitsui Chemicals (4183), KOSE (4922), Nippon Paint Holdings (4612), Unicharm (8113), Shin-Etsu Chemical (4063)
Fabricated Metals
LIXIL (5938), SUMCO (3436)
Applied to Index Forecast
As the graph shows, the human flow data and the industrial production index are generally correlated. The industrial production index, published monthly by METI, is divided into capital volume and labor volume, while labor volume is divided into total actual labor hours and number of employees. In other words, monitoring human flow data at factories can capture the amount of labor and can be presumed to be reflected in the industrial production index.
From these results, it can be expected that the use of daily human flow data can reflect the real situation and provide more accurate forecasts than those published by METI. In the next article, we will define a model that mixes human flow data with the explanatory variables for the forecast and actually make the forecast. The usefulness of the human flow data will be verified by comparing the published forecasts with preliminary and actual data in terms of errors.
Industrial Production Index Forecasted with KDDI’s Human Flow Data (2) is coming soon.
KDDI Location Data on QUICK Data Factory
https://corporate.quick.co.jp/data-factory/en/product/data038/
Follow us on LinkedIn to stay updated with Japan Markets.
